Exercise during growth seems to build a stronger skeleton resulting in a high peak bone mineral density (BMD) in men. Exercise during adulthood produces benefits in BMD or prevents bone loss, but the changes are of minor biological significance as regards fracture reduction. However, prospective intervention studies suggest exercise to improve muscle strength, co-ordination and balance, even in octogenarians—all traits possible to reduce the number of falls. It is virtually impossible to undertake a randomized blinded study of exercise with fracture as end point due to the large cohorts needed. Retrospective and prospective observational and case control studies suggest activity to be associated with reduced fracture risk. This may be correct, but consistently replicated sampling bias may produce the same observation. The Achilles heel of exercise is the cessation of physical activity. Biologically important benefits in BMD or improvement in muscle size and strength achieved by exercise during growth and young adulthood seem to be eroded in retirement, leaving virtually no remaining benefits in old age, the period when fragility fractures exponentially rise. On the contrary, continued exercise on a lower level may maintain some of the musculoskeletal benefit, but dose– response relationships need to quantified, as do the effects of exercise on bone size, shape, architecture and frequency of injurious falls. Absence of evidence is not evidence of absence of effect, but if we recommend exercise then should this be to children, adults, elderly men or men with fractures? What type of exercise? For how long? How many fewer fractures will result in the community from a community based exercise campaign like the anti-tobacco campaign? The higher level of proof, suggesting exercise to reduce spine and hip fractures must come from well designed and executed prospective randomized studies. Blinded studies obviously cannot be done but open trials can, and should be undertaken.